Strings starting with "{%" inside code blocks are parsed as unclosed Markdoc, clogging builds about markdoc HOT 4 CLOSED

By default, the Markdoc parser attempts to parse Markdoc tags that appear inside of fenced code blocks. This behavior can be disabled individually per fenced code block by putting a process=false annotation in the fenced block's info line:

```js {% process=false %}
// Things that look like tags are ignored
console.log("{%"); 
```

We probably ought to add a global configuration option for this so that it doesn't need to be set individually for each fence.

from markdoc.

We probably ought to add a global configuration option for this so that it doesn't need to be set individually for each fence.

That would be great! I would also vote for making process=false the default.

from markdoc.

{% process=false %} created some issues with my highlighter (prism-react-renderer), specifically crashing because it was trying to process a code block that was undefined? Seems like it just skips over it. If I try to call the Fence component like a custom component ({% fence process=false %}) then it seems to just skip over processing the entire block, so that clearly isn't the way. Anyway, for now I've just escaped the strings.

from markdoc.

I had some Liquid template code I needed to render as a code block and given the similarities of Liquid and Markdoc, the page wasn't rendering (non-closed tags and many other issues as expected). I tried the {% process=false %} method which perhaps combined with my Prism highlighter (Next.js) fixed the errors, but also skipped the entire code block (and empty element was returned).

I managed to fix the issue by using the Markdoc's own Doc repository and the markdoc-example tag (https://github.com/markdoc/docs/blob/main/markdoc/tags/markdoc-example.markdoc.js).

Your milage may vary, but it worked for me.

from markdoc.